Github user marmbrus commented on the pull request:
https://github.com/apache/spark/pull/5475#issuecomment-91931540
Thanks for working on this and for adding a test! I have two issues with
the way this is being implemented:
- We are still leaking accumulators on a per table scan basis, and so you
are still going to have a problem if you never explicitly uncache the table
- It seems like we should be able to do this without adding a global hash
map. In particular, when you are uncaching you have a handle to the
InMemoryRelation. Perhaps that class should define an `uncache` method on this
class that takes care of both `unpersist`ing the data and unregistering the
accumulators.
The ones inside of the table scan are clearly harder to deal with, but
perhaps we can just make those lazy vals that are only initialized during
testing due to some conf being set?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
